Opsc 16690 #17
Draft
Opsc 16690 #17
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Fixes regression in launcher that prevents running from bin directory.
…rous tests. Issue is kept open to discuss better future solutions.
…ibs where an update attempt caused notable issues were left unmodified.
This removes an obsolete defence against encoding errors during the processing of exceptions, which is now dealt with in Py.dispayException.
Problem was a missing assignment to the message variable. Also brought words closer to CPython 2.7.13.
…s._args_from_interpreter_flags.
These functions now match CPython expectations. In many cases, the root cause was in _socket.py.
This partially backs out cffe5b824a21, to restore synchronisation, but keeps the IOD singleton pattern to protect the type registry. It backs out f3458ef83e08, now unnecessary with PyType synchronised. See also discussion in #2487.
This was a regresion caused when trying to deal with non-ascii paths in the preparation of a log message. Thanks to James Mudd for diagnosing this. Also jython#86.
We leave exiting to the JVM when a main Python program was run and ended with good status.
Where a unicode argument is given, it is interpreted as bytes through the default encoding. A test suite is added (missing from CPython 2) to extend test_binascii to unicode arguments (ASCII decoding only). There is some tidying up in binascii.java, but more is needed.
There are no functional changes, but some re-ordering and renaming, and some dead statics are removed. The choice of data types is reconsidered to reduce masking and casting.
A change to PrePy.getJarFileNameFromURL(URL) in cset:159c277c4a80 broke the test. getJarFileNameFromURL now no longer insists on a particular class as target (so we can avoid e.g. waking the type system). Also, let's test it with non-ascii file names.
This is to investigate issue #2834, but is not a fix. It provides a failing test (except that the failing assertions are commented out). These assert that importing the same class definition to interpreters with distinct class loaders results in distinct Python type objects and Java classes. In fact, the first interpreter ends up sharing them with all subsequent ones through the common package manager.
Part of the investigation of #2834, adds use of Thread.sleep to some of the test scripts, access to which fails quite frequently. The case testSeparateLoader2() does not produce failures because incorrect behaviour is currently not tested.
It is necessary to allow PyType.fromClass() to return incompletely consructed types when building dependent PyTypes in a single thread. This change intends they should not escape to other threads. We use a RuntimeException to return a PyType without binding it to its class prematurely. Succinct debug methods are left in the code (disabled). We've had to write these several times to address similar bugs.
CVE-2013-2027 points out that Jython may be run with umask 0, and then files cached will be world-writable affecting later sessions. #2044 claimed this fixed by other work, but this change fixes the permissions explicitly in the compiler and package manager.
This change causes the default engine context in a PyScriptEngine to be created as a module __main__, and __name__ to be defined, comparable to the environment found when executing a Python program. The thread-local name space concept is otherwise undisturbed. org.python.jsr223.ScriptEngineTest cases are modified correspondingly, and testThreadLocalBindings is fixed to test isolation within a thread. --HG-- extra : amend_source : 936c178b0add1449e2d1d83a4d2161cd5bcd0ce0
This change updates lib2to3 tests from the CPython stdlib, skipping where necessary. Rather than call out to the shell for diff (fails on Windows) we use a comparison function in Python, being careful to choose the mode rbU to deal with CRLF line endings.
This change addresses a test failure observed in test_jython_initializer where the build Java is ahead of the default installed Java. We "inherit" java.home in the otherwise-stripped environment. --HG-- extra : amend_source : b595dda0ceda305384b4b9fb5cd2f9a84af79df1
We allow up to 2 seconds discrepancy in the logic that decides whether to re-compile .py source to $py.class files. This is to accommodate the rounding of last-modified time that occurs when storing and extracting files from a JAR, as during installation. We choose to round times down when JARring and allow 2 seconds positive difference in the check. This has benefits beyond #2862, in avoiding spurious recompilation, so is made a distinct commit. At the same time the implications of #2862 are arguably not addressed fully until the JAR cache is made user-local. --HG-- extra : amend_source : b01f31c727008f16808e4095db542f9af3949967
We no longer try to cache every user's JARs in the installation directory, ineffective since fix of #2044. Behaviour for applications that set an absolute location (or skip) is unchanged.
If the map was modified during a call to keys(), the keyset would change size and cause an ArrayIndexOutOfBoundsException. This creates a copy of the keyset and uses that throughout the method. --HG-- extra : amend_source : 025dd98595b444af0c4ffb67ae4bd4252fb752b4
Shading the Bouncy Castle classes in our JAR makes them untrusted as a security provider. We document that to work around this, users provide the JARs directly.
Update version numbers. Add jffi-dll-droppings to .hgignore so as to avoid complaints from the build after test.
... whilst hoping the next release is actually 2.7.2
…o we can know if they failed
Fix options so they actually work, set hard defaults
The commit includes changes to Build.xml to version the jython-standalone jar file.The commit also includes a Jenkinsfile which will be used to build the project with ant and publish the artifacts to artifactory.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.